Larry the Cow ...

[dwisp]

Hi all,

I've got several questions about the process of developing gentoo.

First of all, I'd like to know how desicions are made within gentoo. Who has how much to say ? I've seen there're many developers, (forum-)administrators, and so on, but who has how many rights deciding what's the furure of gentoo ?

Anther point : Given I'd be writing a hardware autodetection on bootup, default (user-friendly) kernel configuration and a graphical installer, how would be the chances (in case those tools did work more or less without problems) that they could be integrated within the 'main' gentoo branch ?

My questions arise due to a discussion here https://forums.gentoo.org/viewtopic.php?t=10887. If you dont want to read the whole thread, my point of view boils down to a few things :

o) I think gentoo is currently one of the (computer-)technically most advanced distributions, but it has not reached it's full potential yet.

o) being user-friendly is an technique too, being user-friendy AND flexible is an art; I think gentoo is predestinated to fulfil it.

o) yes, I *do* want to hide away any commandline prompt, any kernel specific issue or anything else not related with the primary tasks of a gentoo-linux-user; but no, I *do not* want to take away any flexibility, or any other way using gentoos powers.

o) some people said, gentoo is (or at least should be) for power users, being able to exploit even the last dark corners of their system. I myself am a poweruser, I'm a linux developer and I've been working with the commandline for several years now, so I know how mighty these tools are and how flexible. But I know too, that in 98% of all cases i just want to do the default action, use the default switch or enter the default value and I really, really dont want the system to bother me with stuff that it could do on it's own.

All in all, I want a system that helps me as much as it can, but respects my own 'thoughts' if it can't.

So please feel free to reply your own thoughts and let us discuss this issue. It is an important one, especially because gentoo has the potential to succeed where other have failed.

[abhishek]

I dont think kleiber or nitro(fourm admins) have more of a say then the rest of us, ocnsidering that they're not devs. I think ultimately the devs decide what goes and what doesnt go into gentoo(could be wrong). Personally, I don't think there should be a graphical installer in the main install, but i htink that has been beaten to death in that other thread, so don't see a point in mentioning it agian.

[Evangelion]

I think there's nothing wrong with graphical installer, assuming that there's still option for text-only install and that the GUI doesn't limit the power & flexibility of the install.

If there is a way to make things easier without sacrificing power, there is IMO no reason not to do it. Making things difficult on purpose is just wrong. I'm not saying that the install is difficult on purpose, but would be wrong not to make it easier even if there are no technical reasons not to do it. Yes, the installation-procedure should be flexible and powerful, but that does not mean that there must not be any kind of GUI or automation. You can have both without sacrificing power of the installer. I think the only thing that supports having difficult installer is the l33t-factor . If you can install Gentoo, you are l33t. Making installer easier would remove some of that l33tness .

[dwisp]

[abhishek]

If you want ot mamke a GUI installer, shoot. /me will be happy with his 1.2 iso and wget. That's one of thae parts I like about a CLI installer. It's real easy to get a newer image then what's on the CD. If you do make it, could you make sure you incorparate that?

[phong]

[klieber]

Moving this to Gentoo Suggestions. This thread already has the beginnings of another "yes, we should have/no we shouldn't have a GUI installer." That's been beaten to death already, so if this thread starts to turn into that too, it's getting locked.

As for who has say about what, drobbins is the project leader. He has the final say about everything. The rest of the organization is very loose -- many decisions are made in #gentoo-dev and on the gentoo-core, a (private) mailing list. There isn't really a formal hierarchical command structure. But, then again, there rarely is in most open-source projects.

--kurt
_________________
The problem with political jokes is that they get elected

[mcking]

[rac]

[masseya]

I think if you are interested in doing something like this you should just start doing it. Even if it's not picked up by gentoo something like a better installer is going to be popular by some distro somewhere. Because there's really no structure in Open Source software you really don't have to ask around and get permission for something like that. I would really like to see the types of things you are talking about. I don't know if I would like them or not, but I would sure like to try them out.
_________________
if i never try anything, i never learn anything..
if i never take a risk, i stay where i am..

[dwisp]

Woohoo, lots of answers till now - thanks to all participating so far Let me go into detail with some of the responses.


The Adminstrative Part

data, yes you're probably right about the local forum admins (thanks klieber , but that was my problem : who's who. I've seen lots of different oppinions among the gentoo staff, so I want(ed) to figure out whether there's a defined path gentoo is already following or if its still open for something "completly different".
Some point I really consider problematic are personal oppinions from officials. If I see an "developer", "administrator" or any other one having five stars due to a special function saying "not needed" it sounds to me like "the gentoo doesn't need it". Please anyone, don't get me wrong, it's just when you're in an official uniform, you are supposed to do an official job, or at least make clear that anything else is your own 2 cents.

The Graphical Installer

I *really* dont want to start another thread in here about an GUI, so please just let me answer some comments and leave it like that.
Evangelion, full ack, your comment says it all. Ehm, data, whats a CLI installer *scratches his head* ?

Other Useful Tools

There's a document called "Top 10 thinks that are wrong with Linux today" which I downloaded some weeks ago (sorry, forgot the url - but google may help you) and read through it several times. So I asked myself : What can be done to fix them ? Some points (e.g. the X-redrawing issue) is far beyond my skills and knowledge about the whole X architecture, but others could easily be done. (If you really want to know what comes my ideas very close go to http://www.knoppix.net and download the iso image. Yes, it's debian based, but the hardware autodetection ... it's amazing. Burn a cd ? Just Works. Sound ? Just Works. Network ? PnP. X ? Autoconfigured ? Connect to the Internet ? My Grandma could (nearly) do it.) Unfortunatly debian itself collides with my sense of art, and it a bit, well, conservative. But gentoo is beautiful. In my eyes primarily because of it's use of the source. Linux spirit is the source, so why not use it ? Indeed, I'd rather say : Linux (linux in sense of GNU/Linux) is implicitly designed to require the source (everyone else loving those 'unresolved symbol ..." messages will probably agree and has major problems in providing a binary stable API across all its "mutations".
Thats why I think gentoo was the highest odds to become some sort of reference implementation people can call "the linux system".
But lets get specific, what tools am I talking about :

o) hardware detection on bootup, at least for the unproblematic devices or on request (like sound, net, cdrw, pcmcia, usb, firewire, mass-storage, printer(?) )

o) an "assimilate this package" feature. This is probably where I'll try to focus (if I should find time somewhere). Does mean ? User d'n'drops an .tar.[bg]z file (or an url to one) into an app which first checks if this file is already within the portage-tree and uses this ebuild or it automagically generates an default one and "assimilates" this package. After that the ebuild is sent somewhere to be quequed for manual fine-tuning. Or, in case of an error, an message with the last n lines will be sent to an list for assistance (and improvement of the program).
[something like ebuilder, but with an gui and fully automatical]

o) easy encrypted directories (which basically means : automatic setup and detection of encrypted loop devices)

o) several autoheal functions like a "repair this app" (in sense of broken libs; isn't here being worked on already ?) , "really, really do update my system" (here too ?), "do update without being tricked by man in the middle", "brew coffe"


The Kernel

Not much to say to this one : rac mentioned CML2, but AFAIK did ER more or less retire from the development. At least my linux-magazine says this. I myself am not a friend of an dynamical kernel config, cause it's very likely that you need just this module you didn't compile. I'd prefer an default .config file with ext, xfs ide and scsi enabled by default and everything else compiled as modules. If you don't care about the kernel this would be enough, if you do, you can switch happily yourself. (this is my ack to mcking, or what I consider the default settings for the kernel


finally{}

Tristam29, as easy as your word sound as difficult it's (for me) to follow them. It would really drag me down if I tried to improve [PutYourNameHere] just to find out my work isn't 31337 enough. But I guess you're right. It just though I'm a pure linux-user for 5 years now and a developer too, I'm very new to the active development within the open source scene. So I was just afraid of working against the gentoo-way and afraid of having to realize that gentoo might not be for what I thought it to be (if it should already be there).



Think thats enough for today, time to go to bed



(changed the smily to fit my current look)

[masseya]